This is one of four identical applications for a multi-center study of long-term treatment strategies for panic disorder (PD). The study builds upon the findings of our original multi-center study comparing imipramine, placebo, cognitive behavior therapy (CBT), and their combination in the treatment of PD patients with no more than mild agoraphobia. That study found response rates were as high with CBT or imipramine alone as with their combination. Given the added cost of combined treatments, it therefore seems reasonable to begin with monotherapy. Further, following general principles of medical practice, it would be reasonable to initiate treatment with the less invasive behavioral intervention. It is then important to learn what should be done following initial treatment. The proposed grant addresses this subject. In the proposed study, all subjects initially receive CBT alone. Subjects are then randomized in one of two postacute studies, depending on response status. These studies address two questions: (1) among patients who respond to an initial course of CBT, is maintenance therapy required to maintain response?; and (2) among those who fail to respond substantially to an initial course of CBT, will the addition of pharmacotherapy improve response? A total of 336 subjects will be enrolled from 4 sites over a 3 year period. Responders will be randomized to a maintenance study comparing no maintenance with 9 months of continued CBT. Non-responders will be randomized to a study comparing paroxetine with continued CBT. In addition to the main questions, possible predictors of response and relapse will be examined as well as possible mediators of response. Central to this proposal is the continuation of a well-functioning and productive collaboration among the sites conducting the current study. These sites already have in place the trained and certified assessment and treatment personnel and rigorous monitoring procedures that are essential for the conduct of a study such as the one we propose.